README.md

metadata

language: en
license: apache-2.0
tags:
  - fine-tuned
  - gemma
  - lora
  - gemma-garage
base_model: google/gemma-3-1b-pt
pipeline_tag: text-generation

test-new

Fine-tuned google/gemma-3-1b-pt model from Gemma Garage

This model contains LoRA adapters fine-tuned using Gemma Garage, a platform for fine-tuning Gemma models with LoRA.

Model Details

• Base Model: google/gemma-3-1b-pt
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Training Platform: Gemma Garage
• Fine-tuned on: 2025-07-26
• Model Type: LoRA Adapters (not merged)

Usage

Option 1: Load with PEFT (Recommended)

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("google/gemma-3-1b-pt")
tokenizer = AutoTokenizer.from_pretrained("LucasFMartins/test-new")

# Load and apply LoRA adapters
model = PeftModel.from_pretrained(base_model, "LucasFMartins/test-new")

# Generate text
inputs = tokenizer("Your prompt here", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Option 2: Merge and Load

from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("google/gemma-3-1b-pt")
tokenizer = AutoTokenizer.from_pretrained("LucasFMartins/test-new")

# Load and merge LoRA adapters
model = PeftModel.from_pretrained(base_model, "LucasFMartins/test-new")
model = model.merge_and_unload()

# Generate text
inputs = tokenizer("Your prompt here", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Training Details

This model was fine-tuned using the Gemma Garage platform with the following configuration:

• Request ID: 43a3a2fd-ada0-40f1-9a29-9f4050d94bcf
• Training completed on: 2025-07-26 18:44:17 UTC

For more information about Gemma Garage, visit our GitHub repository.